I'm want to run tensorflow on a very standard machine setup (windows 64 bit) and have read that tensorflow has greater performance if built from source as it is optimised for your system. When installing tensorflow via pip for why does pip not select the optimal build for your system?
Also if you did install via pip is there a way or being able to tell whether the optimal build has been installed, or is the only way of knowing that simply remembering how you installed it?
Google has taken the position, that it is not reasonable to build TF for every possible instruction set out there. They only release generic builds for Linux, Mac and Windows. You must build from source if you want all the optimizations for your particular machine's instruction set.
Related
I want to run TensorFlow on my microserver. I'd like to install a non-systemd Linux if possible e.g. Alpine, but I am new to TensorFlow and I am not sure how much it relies on systemd or if it would run without it. Would it?
TensorFlow is a regular programming library, it's not a system library and not running as a service, thus it isn't dependent on systemd.
I've tested TensorFlow on Windows 10 Subsystem for Linux (WSL) which doesn't come with systemd and it's still working.
I'm running a Windows computer with just a CPU (no GPU). When I run pip install tensorflow -vvv in order to see what pip is doing, it lists a lot of links, but for all of them, it says "Skipping link ... it is not compatible with this Python."
Does tensorflow support Python 3.6.4 on Windows? If so, what binary URL should I use to install it?
(I previously installed with this version due to reading this, but ran into this error without the DLL load failed message, so I'm wondering if there's a better version I should use.)
Also, I'm aware that Tensorflow says they support Python 3.x, but right now it hasn't been working for me.
You have probably installed Python 32bits, you need the 64bits version
Whenever I use tensorflow, it displays the message "The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available on your machine and could speed up CPU computations" and 2 more similar messages.
So I decided to build tensorflow from source to get rid of these messages. I'm using python 3.5 on Debian, and followed the instructions at https://www.tensorflow.org/install/install_sources (CPU only, no GPU).
It asked during the build if the build should be for the machine it's doing the build on, I selected that, it included -march=native in some compiler option.
Everything seemed to work, but when I ran python3 to test the build, it still gives the messages about "The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available ..." etc. How to I make the build use the hardware that it's running on?
There have been similar questions, and most of the answers to them are wrong. They say it's necessary to specify options like "--copt=-msse4.1 --copt=-msse4.2" in the build; it isn't. With the default option "-march=native", the GNU compiler will use SSE4.1 and SSE4.2 instructions if they are available.
The real problem is that if you build tensorflow from source, after installing the default build with pip, pip won't replace the old build with the new build. Everything will seem to work, but your old build remains in place in a directory under ~/.local.
The solution is simply to uninstall the old tensorflow with pip ('pip uninstall tensorflow' or 'pip3 uninstall tensorflow'), and then rebuild from source. If you have already done a build, and wondered why nothing seemed to change, you needn't repeat the build but can just execute the last couple of steps (https://www.tensorflow.org/install/install_sources), namely bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg , followed by the pip install.
I am looking into using Tensorflow for my research soon, and looked at the online documentation for installing with Conda https://www.tensorflow.org/versions/r0.11/get_started/os_setup.html#anaconda-installation.
It suggested creating a new environment, and installing Tensorflow in it, and the installing other python packages afterwards.
But I already have an existing environment with lots of packages I need, and I'm wondering if its safe to add Tensorflow into that environment?
Also, I have a question about how this installation with conda works. I know that Conda will create a distinct set of folders containing the libraries needed for each environment, but if I install Tensorflow, what happens to all the base low level C++ and CUDA libraries that get compiled? Do they reside in my Conda environment's folder, or are they in some system wide libraries closer to my root?
PS: I'm using Ubuntu 16.04, and have a GPU that I want to run Tensorflow on.
Thank you.
But I already have an existing environment with lots of packages I need, and I'm wondering if its safe to add Tensorflow into that environment?
conda has this awesome feature called "revisions". You can show your current environment with
conda list --revisions
which allows you to revert changes to your conda environment. This allows you to install new packages with confidence that if something breaks you can always revert it later. See this page for more info: https://www.continuum.io/blog/developer/advanced-features-conda-part-2. tl;dr: conda install --revisions <revision_number>
what happens to all the base low level C++ and CUDA libraries that get compiled
Are you talking about the libraries that get compiled when you are trying to run your code? Or the C++/CUDA libraries? If you're talking about the C++/CUDA libs then conda is not compiling them, but merely installing a pre-compiled binary into a specific location that gets picked up. If you're talking about your code, then where exactly those files live would seem to depend on where you put them.
So the documentation mentions that it is better to install from source, then build a pip package. Why is this recommended over doing a direct pip install using the wheel file provided on the downloads page? here
I tried the direct pip install and then runnning some scripts in the inception folder.This results in errors with bazel not finding some of the dependencies. I am guessing this is related to not building tensorflow from source, but I can't figure out why this is the case. Any pointers? Thanks!
Installing from pip is supported, can you provide more details on your os and the specific errors you saw?
The main reason to build from source is simply performance.
Building and installing from source
The default TensorFlow binaries target the broadest range of hardware to make TensorFlow accessible to everyone. If using CPUs for training or inference, it is recommended to compile TensorFlow with all of the optimizations available for the CPU in use. Speedups for training and inference on CPU are documented below in Comparing compiler optimizations.
To install the most optimized version of TensorFlow, build and install from source. If there is a need to build TensorFlow on a platform that has different hardware than the target, then cross-compile with the highest optimizations for the target platform. The following command is an example of using bazel to compile for a specific platform
ref: https://www.tensorflow.org/performance/performance_guide